WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 11 Models and experimental results from LQ, HQ ( and more) and QXF Giorgio Ambrosio Fermilab WAMSDO CERN January 16, 2013 With contributions by: Helene Felice, Shlomo Caspi, Tiina Salmi, Maxim Martchevsky (LBNL) Guram Chlachidize, Linda Imbasciati (FNAL) Massimo Sorbi, Lidia Rossi, Vittorio Marinozzi (Univ. of Milan) Paolo Ferracin, Ezio Todesco, Marta Baiko, Hugo Bajas (CERN) Giulio Manfreda (Univ. of Udine) WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 22 Outline What is the maximum acceptable temperature at the hot spot in Nb 3 Sn accelerator magnets? What feedback from magnet test to QP codes? Where does the QXF stand? WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 33 MAXIMUM ACCEPTABLE TEMPERATURE AT HOT SPOT? In Nb 3 Sn accelerator magnets WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 44 High Hot-Spot Temperature Test in Quad Test performed on TQS01c (1m quad) with MJR conductor (47% copper) Spontaneous quenches (pole turn, inner layer) same segment during all study; High MIITs (and Temp) by removing protection features Iq ~ 80% ssl at start Iq_max: + 4% Iq_min: - 25% +3.3% -7.2% +4% -25% -7.4% Fermilab TD Note: TD : LARP TQS01c Test Summary G. Ambrosio, R. Carcagno, S. Caspi, G. Chlachidze, F. Lewis, A. Lietzke, D. Orris, Y. Pischalnikov, G.L. Sabbi, D. Shpakov, C. Sylvester, M. Tartaglia, J.C. Tompkins, G. Velev, A.V. Zlobin WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 55 Fig. 9. Epoxy de-lamination and slight inward cable displacements on one side of coil-15s inner-pole island after high-MIITs study. IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 18, NO. 2, JUNE Test and Analysis of Technology Quadrupole Shell (TQS) Magnet Models for LARP S. Caspi, G. Ambrosio, A. N. Andreev, E. Barzi, R. Bossert, D. R. Dietderich, P. Ferracin, A. Ghosh, A. R. Hafalia, V. V. Kashikhin, A. F. Lietzke, I. Novitski, G. L. Sabbi, and A. V. Zlobin WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 66 Hot Spot Temperature Computation Hot spot temperature computed with QuenchPro: MIITs vs. Temperature including epoxy and insulation in enthalpy computation Adiabatic approximation Assuming peak field on cable (at quench current) Constant in QuenchPro RRR: (range of RRR in quenching coil) RRR of quenching segment not available WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 77 Hot Spot Temperature vs. Degradation All temperatures are in K +/- 6 K (for RRR uncertainty) WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 88 Tests on Cable Samples and Small RacetrackQuench Protection Issues of Nb3Sn Superconducting Magnets for Particle Accelerators L. Imbasciati, PhD dissertation Fig. 6.8: Summary of quench experiments: reduced current (quench current divided by maximum current) vs. peak temperatures reached during the preceding quench test. The lines represent the temporary sequence of the peak temperature events. Peak temperatures measured by resistance growth with voltage taps around hot spot Degradation of electrical strength WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 99 Conclusions - I T_hot spot > T 1 Active territory: Hot spot after quench is not in the same strain/stress state where it was before the quench Magnet may train, detrain, effect is reversible T_hot spot > T 2 > T 1 Degradation territory: Degradation is irreversible and/or the magnet may experience insulation failures T1 = K based on TQS01C T1 > 400 K based on results in L. Imbasciati dissertation Glass Transition temperature of CTD101K = 386 K This may be the physical limit! AIP Conf. Proc. 614, pp ; Highly radiation-resistant vacuum impregnation resin systems for fusion magnet insulation P. E. Fabian, N. A. Munshi, and R. J. Denis and CTD-101K Material datasheet Max acceptable temperature = 386 K - margin WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 10 Conclusions - II Fig. 6.6: Quench integral accumulated during the quench experiments performed on cables (above), and during the small magnet experiment (below), compared to curves calculated using the heat balance equation including metal components only, with epoxy resin and with 0.15 mm insulation. Quench Protection Issues of Nb3Sn Superconducting Magnets for Particle Accelerators L. Imbasciati with RRR = 70 the hot spot temperature changes 340 K 383 K Computations may slightly underestimate the hot spot temperature because: They include epoxy and insulation They compute cable average temperature Local RRR may be higher (for instance at cable edge) Material prop. may not be correct (for instance G10) WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 11 Conclusions - III This picture may change if another material is used for potting: The glass transition temperature may change Other mechanisms may cause detraining or degradation These tests should be repeated on magnet with cored cable Core and cable may not come back to the same condition after quenches at high temperature, WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 12 FEEDBACK FROM LQ TEST WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 13 Long Quadrupole Main Features: Aperture: 90 mm magnet length: 3.7 m LQ Design Report available online at: https://plone4.fnal.gov/P1/USLARP/MagnetRD/longquad/LQ_DR.pdf ParameterUnitLQ N of layers-2 N of turns-136 Coil area (Cu + nonCu)cm Coil Lengthm3.3 LQS01 SSL4.5 K Current13.7 kA Gradient240 T/m Peak Field12.25 T Stored Energy460 kJ/m Long Quadrupole Overview G. Ambrosio 14 LARP Collab. Mtg 10 Port Jefferson, Apr , 2008 Quench Protection Goal: MIITs < 7.5 Temp ~ K (adiabatic approx) Quench protection param. (4.5 K) conservative hypothesis Dump resistance: 60 m (extract ~1/3 of the energy; V leads ~ 800 V) 100% heater coverage ( heaters also on the inner layer) Detection time: ~5 msbased on TQs with I > 80% ssl Heater delay time: 12 ms based on TQs with I > 80% ssl 6 ms (transv. propagation through insul.) + 6 ms (long. propagation btw heating station) Very challenging! J in copper = 2900 A/mm 2 at 13.9 kA (4.3 K SSL) WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 15 Measurement vs. Computation Tests showed that there is margin: MIITs lower than computed Faster current decay, With one exception: quench in midplane block at 11.3 kA RRR higher than value used in computations. Hot spot temperature lower than computed values COMPUTEDMEASURED Current (A)MIITsRRRTemp (k) MIITsRRRTemp (k) < 247 WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 16 Feedback from LQ test Current decay faster than computedStudy of Superconducting-to-Resistive Transition for the US-LARP Large High Field Quadrupoles for the LHC Upgrade Lidia Rossi, Bachelor dissertation At very start! 0 10 ms WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 17 Time constant at current decay start Time constant at decay start: = L / R with R = R dump + R busbars (R coil is negligeable) measured < estimated (240 ms) measured used to evaluate L effective WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 18 Dynamic Inductance vs. Measurements L effective < L dynamic Large variation of L with frequency at room temperature reduction of L effective due to eddy currents (cable, structure) Dynamic inductance computed by OPERA WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 19 New Comparison Better modeling after including these and other improvements into QuenchPro WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 20 Quench-back at High Current? Time constant at decay start: = L / R At I q /I ssl > 88% measured becomes smaller Quench back? Multipole quenches due to current redistribution? Magnet sitting at a constant current: from 5 to 13 kA (NO quench) Discharge in the 40 m dump resistor without PH Does the magnet quench from eddy current generation in the cable (form of quench- back)? From the current decay: At 5 and 10 kA: no sign of quench A 13 kA: signs of quench At 15 kA: fraction of the magnet is quenching Last 15 kA test with PH: no clear impact HQ01e Quench-Back 2012/11/14 2nd Joint HiLumi LHC - LARP Annual meeting 21 H. Bajas et al., Test Results of the LARP HQ01 Nb 3 Sn quadrupole magnet at 1.9 K, presented at ASC2012 HQ01e at CERN WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 22 Conclusions These features may help the protection, but should be well understood: Above what I q /I ssl do they have a significant effect? Effect of cored cable? Note: maybe a dump resistor may help to trigger them even if the energy extracted is no so significant Generally speaking: do validation of QP codes with real data! WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 23 LQ Coils after Test Delamination on Inner layer Heater coil Insulation heater Insulation coil Also one heater-coil short Possible causes: Superfluid helium + quench Seen in TQ coils Heat from heaters on ID Not done in TQ coils Options: Strengthen insulation Not good for cooling Change heater location Best solution 23 WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 24 Quench Propagation Velocity Measurements at 4.3 K IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 17, NO. 2, JUNE 2007 Assembly and Tests of SQ02, a Nb3Sn Racetrack Quadrupole Magnet for LARP P. Ferracin, G. Ambrosio, E. Barzi, S. Caspi, D. R. Dietderich, S. Feher, S. A. Gourlay, A. R. Hafalia, C. R. Hannaford, J. Lizarazo, A. F. Lietzke, A. D. McInturff, G. L. Sabbi, and A. V. Zlobin WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 25 QXF PROTECTION (preliminary results) By Massimo Sorbi, Giulio Manfreda, Vittorio Marinozzi WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 26 ROXIE QLASA Comparison Comparison btw QLASA and ROXIE using same material properties (MATPRO library) and assumptions: Magnet length: 8.5 m Total stored Energy: 15.2 MJ for Qlasa 12.2 MJ for Roxie, with the inductance variation Rdump = 58 mohm Vmax = 1000 V (+/- 500 V) Heaters on Inner and Outer layers WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 27 Hot Spot Temperature vs. Delay Time 370 Whole magnet should be quenched in less than 45 ms! WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 28 Can we achieve this delay time with heaters only on outer layer? Delay time after detection = Validation time + Switch time + Heater-coil diffusion time + Longitudinal propagation btw heating stations All Outer Layer quenched Layer-layer diffusion + Longitudinal propagation Whole magnet quenched IL OL __ In order to have all magnet quenched in 45 ms: WAMSDO, January 16, CERNModels and experimental results G. Ambrosio 29 Preliminary Conclusions We should calibrate ROXIE and QLASA with experimental results We should test Nb 3 Sn magnets with cored cables at high hot spot temperatures We have to fight for every ms in the design of MQXF and its protection system Together we will make it!